Reservoir filling: implications for reservoir geochemistry

Cartoon sketch of example where density stacks within a reservoir create a vertical / lateral grading.
Cartoon sketch of example where density stacks within a reservoir create a vertical / lateral grading.

This short technical note discusses concepts around how petroleum reservoirs are thought to fill and how that process impacts the understanding of reservoir fluid’s geochemistry and PVT properties such that these tools can be deployed effectively. Enhanced understanding of the fluid properties offers many opportunities for insightful and low-cost approaches to field management and problem solving, from compartmentalisation studies to production allocation and monitoring.

While geologists spend much time and effort on the static description of a reservoir (poro-perm, continuity, structural heterogeneity), the fluid description is often treated simply and only considered at the point of sanction. This almost certainly means that opportunities for enhanced understanding and more effective management and exploitation strategies are missed. If deployed appropriately, reservoir geochemistry can be a flexible, cost-effective tool to help with the parameterisation and management of petroleum reservoirs. As a discipline, reservoir geochemistry aims to exploit identifiable heterogeneity in fluid geochemistry to provide insight to its subsurface continuity and physical properties; for example, most if not all oil fields have gradients in bulk properties such as API and GOR. These properties are crucial to valuing a petroleum asset and mapping their variation will provide insight to the reservoir system. As a concept, reservoir geochemistry is certainly not new, many of the ideas and concepts were available in the 1980s and 1990s; Larter & Aplin (1995) were perhaps the first to draw together a holistic framework for the discipline. Despite providing a direct assessment of the fluids of primary interest in any field exploitation scheme, at very low-cost, the take up of reservoir geochemistry as a discipline has been at best modest. This is likely to reflect a range of issues, from knowledge within field development teams to perceptions of the tools as unproven or superfluous.

While uncertainties remain, Stainforth (2004) set out a useful model for reservoir filling which is perhaps not widely appreciated. It is commonly assumed that petroleum charge homogenizes within the trap, and thus compositional gradients reflect the physical forces the petroleum column is subject to (gravity, temperature, diffusion, etc). However, Stainforth (2004) argues that this would entail an unfavourable generation of potential energy. In his paper he considers two end-member models: one where petroleum charge mixes perfectly within the trap and a second where it does not mix at all. These end-member models are considered further in this note.

Integrating PVT data is a crucial aspect of any reservoir geochemistry study. Saturation pressures (Pb = bubble point; Pd = dew point) have a strong influence on production behaviour, however these parameters are often treated in a simple way i.e.

  • Values are constant in depth (Figure 1A)
  • Values are equal to reservoir pressures (Figure 1B)

Field data rarely conform to such assumptions and depth plots are often curved (Stainforth 2004 and refs. therein), only converging at the GOC (Figure 1C &D), indicating that petroleum fields are only in local thermodynamic equilibrium near the contact, with the degree of undersaturation potentially decreasing so rapidly that bubble point decreases with depth in an absolute sense (Figure 1D).

Figure 1. Fluid pressure (Pf) and saturation pressure trends versus depth in reservoirs with a vapour and liquid leg. (A) assumed constant bubble (Pb) and dew (Pd) points; (B) assumed linear saturation pressures equal to Pf at all depths; (C) observed concave up trends of Pb and Pd; observed concave down Pb and Pd trends. Re-drawn from Stainforth (2004).

Reservoir geochemistry studies often start with the assertion that a field is homogenous and connected. The geochemistry of samples is quantified and differences above analytical error are used to invoke compartmentalisation. In the oil industry composition grading, if considered at all, is commonly assumed to only occur in long petroleum columns and occurs as a result of gravitational forces. Neither of those assumptions is true. Grading is probably ubiquitous to petroleum columns irrespective of phase (cf. Figure 2; Figures 5-9 in Stainforth, 2004). Thus, caution is required since comparing a sample from a shallow point in the column with a deep one may incorrectly invoke a barrier. What we typically observe is:

  • In stacked, discrete oil reservoirs, trends of increasing API, GOR and Pb with increasing depth
  • Within oil reservoirs, trends of an opposite sense, where API, GOR and Pb tend to decrease with depth. Stainforth (2004) interprets these vertical trends as the vertical components of lateral trends – present due to the aspect ratios of most petroleum fields and the inability of lateral diffusion to homogenize the oil in the time since charge (cf. Figure 3).

Stainforth (2004) proposed two end member models that account for the occurrence of grading in petroleum columns:

  • Model 1: Charge – instantaneous, perfect mixing (via diffusion and/or convection?) →gravity & temperature induced grading (or biodegradation induced)
  • Model 2: Charge – no mixing → stacking according to density (Figure 4) → diffusion towards less graded columns in equilibrium with temperature and gravity fields (relatively ‘fast’ vertically, geologically slow laterally); where columns do mix they will do so as a result of gravitational and kinetic processes (Figure 5).

Figure 2. Example of API and GOR grading from a carbonate reservoir in the Middle East. The interpretation of such trends would typically include integrating pressure, petrophysical and geological data.

Figure 3. Cartoon sketch of example where density stacks within a reservoir create a vertical / lateral grading. Lateral length scales are typically at least an of order of magnitude greater than vertical length scales.

A key process in the mixing and homogenising of a petroleum accumulation invoked by Model 1 is diffusion; however, this process has been shown to be typically inadequate to entirely homogenise petroleum fields (e.g. England et al. 1987) reflecting the diffusion coefficients relative to the length scales involved: a field will typically have a lateral extent (usually kilometres) at least an order of magnitude greater than its vertical extent (typically tens of metres) which may be up to an order of magnitude greater than the reservoir thickness. Diffusion is only likely to be of importance in thick homogeneous reservoirs with long residence times. Thus, the fundamental process underpinning Model 1 is likely to be significant within a vertical section (Figure 5) but insignificant laterally in most geological situations.

By considering how fields fill, we may arrive at a better model to explain the common occurrence of graded petroleum columns. In most, but not all, scenarios petroleum charge will become progressively lighter (lower density) with time as the basin is buried and the kitchen increases in temperature. Fluid migrates along reservoir-seal interfaces, once it enters the trap, since diffusive mixing is slow, the most recently arrived (lightest) petroleum will advect through the existing petroleum, as ganglia, to the crest of the trap, displacing heavier petroleum downwards (Figure 4). Diffusive mixing will slowly evolve this stacked accumulation towards equilibrium with the temperature and gravity field. This model is opposite to that proposed in Model 1 where a homogenous but unstable column evolves to a stable equilibrium condition. The subtlety here is that in Model 1 Stainforth assumes a perhaps common belief that reservoirs fill initially from the top downwards such that the youngest petroleum is initially at the base of the column; his argument being that since it is likely the lowest density (highest GOR) this would be unstable. In Stainforth’s model fields stack downwards such that the youngest petroleum advects to the top and then displaces downwards such that the oldest petroleum is at the base of the column – which will then mix over geological time.

Trap and seal configurations can add complications that can preferentially retain the heavier components (e.g. spill) or lighter components (seal leakage). API gradients reflect changes in charge composition. GOR gradients according to Stainforth (2004) predominantly reflect the PT history of the trap; although we believe charge GOR will also play a role.

Figure 4. A model of reservoir filling with no mixing (cf. Stainforth, 2004). Either a continuous or pulsed stream of charge stacks into a reservoir in density order, with lighter (later) charge advecting to the top of the trap (or GOC). Pb trends (and Pd) will be concave down in this scenario (Figure 1D). With enough time diffusive mixing will lead to less graded columns that are in equilibrium with the temperature and gravity fields. Inset from Singh et al. 2016.

Figure 5. Gravitational and kinetic mixing in petroleum columns (slide courtesy of Brian Moffat, Petrophase).

So, what are the implications for reservoir geochemistry?

The starting point of any reservoir geochemistry study is always a review of the available samples, their quality and distribution within the geological framework of the field under evaluation. Contamination can often be an issue and needs careful consideration. The possibility of access future samples will be determined by the current field development plan and the nature of the production facility (wet vs dry trees).

Early recognition of compartmentalisation is crucial, if it is to make a material impact on development planning, therefore an appropriate level of sampling should be considered through the appraisal stage such that the techniques discussed here can be deployed. This requires early engagement with multidisciplinary teams.

Most of the discussion in this note has related to bulk properties such as API and GOR, whereas reservoir geochemistry may use a range of compounds from gases (C1-C5) to biomarkers (C20+). The bulk physical properties such as saturation pressure, of interest to engineers, are strongly controlled by the gas and liquid composition. Engineers will split the composition into (pseudo)component groups. What may be a life’s worth of a geochemist’s research is unceremoniously lumped into a C15+ or even a C6+ fraction! This is entirely appropriate since it allows engineers to model a simplified composition efficiently in order to to answer the questions they are asked. The challenge to geochemists in reservoir studies is two-fold. Firstly, they need to understand what engineers do, what challenges they face and how geochemistry may shine light where standard engineering practices cannot. Secondly, the geochemist must decide what degree of difference in their geochemical data is meaningful, with the (ubiquitous) presence of grading further complicating matters. However, because geochemical studies will typically assess individual compound classes that may have a narrow range of molecular weights (if only gasolines or biomarkers are focused on for example), since they segregate little (cf. Kaufman et al. 1990) compared with bulk properties, trends may not be in agreement nor yield equivalent information on connectivity. To make inferences about connectivity both need to be considered; for example, Westrich et al. (1996, 1999) studied the Bullwinkle Field (Gulf of Mexico) which shows very strong compositional gradients but homogeneous geochemical fingerprints indicating the presence of a graded column in a connected reservoir. Both observations have significance for reservoir effective and efficient exploitation.

Reservoir geochemistry studies should always integrate bulk (PVT) properties together with the geochemical data. The geochemical data need to be viewed with respect to the geometry and length scales of the reservoir and an estimate of the time it would take to homogenize the oil by convection/diffusion versus the time of charging the trap (e.g. Smalley & Hale, 1996).

APT has significant expertise in reservoir geochemistry and relevant experience in both the service and operator environment to effectively design and deploy impactful studies. Key to success is clear framing of the problem to be solved, well thought out sample collection, data integration, data quality and execution aligned with the field development plan.

For more information contact:

Acknowledgments: Thanks to Richard Patience and Mark Bastow of APT AS and Brian Moffatt of Petrophase whose comments are gratefully acknowledged.


England, W.A., Mackenzie, A.S., Mann, D., & Quigley, T.M. 1987. The movement and entrapment of petroleum fluids in the subsurface. Journal of the Geological Society of London, 144, pp. 327 – 347.

Kaufmann, R.L., Ahmed, A.S., & Elsinger, R.J. 1990. Gas chromatography as a development and production tool for fingerprinting oils from individual reservoirs: applications in the Gulf of Mexico. In: Schumaker, D. & Perkins, B.F. (eds) Proceedings of the 9th Annual Research Conference. Society of Economic Paleontologists and Mineralogists, October 1990, New Orleans, pp. 263 – 282.

Larter, S.R. & Aplin, A.C. 1995. Reservoir geochemistry: methods, applications and opportunities. In: Cubitt, J.M. & England, W.A. (eds) 1995. The Geochemistry of Reservoirs, Geological Society Special Publication, No. 86, pp. 5 – 32.

Singh, K., Bijeljic, B., & Blunt, M.J. 2016. Imaging of oil layers, curvature and contact angle in a mixed-wet and a water-wet carbonate rock. Water Resources Research, 52, pp. 1716 – 1728.

Stainforth, J.G. 2004. New insights into reservoir filling and mixing processes. In: Cubitt, J.M., England, W.A. & Larter, S.R. 2004. Understanding Petroleum Reservoirs: towards an Integrated Reservoir Engineering and Geochemical Approach. Geological Society Special Publication, No. 237, pp. 155 – 132.

Smalley, P.C., & Hale, N.A. 1996. Early identification of reservoir compartmentalization by combing a range of conventional and novel data types. Society of Petroleum Engineers, Paper No. 30533.

Westrich, J. T., Knigge, P. O., Fuex, A. N. & Halpern, H. I. 1996. Evaluating reservoir architecture in the northern Gulf of Mexico using oil and gas chemistry. Society of Petroleum Engineers, Paper No. 36541.

Westrich, J. T., Fuex, A. N., O'Neal, P. M. & Halpern, H. I. 1999. Evaluating reservoir architecture in the Northern Gulf of Mexico with oil and gas chemistry. Society of Petroleum Engineers, Paper No. 59518.